Following the official launch of Meta Platforms, Inc.'s (META.US) proprietary AI model, Muse Spark, exchange-traded funds (ETFs) with significant holdings in the company have captured market attention. The Muse Spark AI model is designed to enhance Meta's overall ecosystem by enabling faster and more advanced AI-driven capabilities within its platform environment. Meta's stock closed up 6.5% on Wednesday, prompting renewed focus on ETFs that maintain high allocations to the stock. In total, more than 650 ETFs collectively hold 337 million shares of Meta, underscoring its substantial influence within the fund industry. Currently, funds with the highest portfolio weightings in Meta are attracting particular interest from investors seeking targeted exposure.
After more than a year of organizational restructuring and massive investment, Meta has delivered a key milestone. On Wednesday, the company officially launched its self-developed AI model, Muse Spark. This model represents the first major achievement since Meta announced the formation of its "Meta Superintelligence Lab" (MSL) in June of last year and is also the first significant AI model released following Meta's approximately $15 billion investment to recruit Scale AI founder Alexandr Wang. In June 2025, Meta invested roughly $15 billion in Scale AI and brought its founder, Alexandr Wang, onboard to lead the newly established MSL. This move was seen as a critical step in Meta's "strategic reset" within the AI race, a time when its capabilities in large language models lagged noticeably behind those of OpenAI, Alphabet (GOOG), and Anthropic.
Subsequently, CEO Mark Zuckerberg initiated a series of aggressive measures: large-scale recruitment of top AI talent; streamlining the organizational structure to improve R&D efficiency; and committing hundreds of billions of dollars to build AI computing infrastructure. Muse Spark is the first major "test case" of this new system. Meta describes the model as "specifically built" for its various products. Zuckerberg has previously stated that while the model does not yet match the most advanced systems from competitors, it will provide the Meta AI virtual assistant with "smarter and faster" capabilities. The company stated the model will utilize content from Instagram, Facebook, and Threads to support more personalized and visually-oriented responses. Meta also indicated its new model has strong potential applications in the healthcare sector, noting it has collaborated with over 1,000 doctors to train the model to generate more detailed answers on topics such as nutrition and exercise. Leveraging this model, Meta AI will also introduce a "shopping mode" to help users compare prices.
According to evaluation results disclosed by Meta, Muse Spark surpassed leading models from Alphabet, OpenAI, and Anthropic in certain reasoning and multimodal capability benchmarks. Notably, this model marks a departure from Meta's previous strategy of championing open-source models like Llama, shifting instead to a closed-source approach where the model's weights and architecture are not publicly released. This pivot is highly symbolic. Whereas Meta previously relied on open-source models to build ecosystem influence, it is now moving towards a more commercial and competition-oriented stance, with future monetization likely through API fees or even subscription models. The industry widely views this as signaling Meta's transition from an "AI evangelist" to an "AI competitor."
Comments